Brain Areas Involved in Perception of Biological Motion E. Grossman, M. Donnelly, R. Price, D. Pickens, V. Morgan, G. Neighbor, and R. Blake Vanderbilt University Abstract & These experiments use functional magnetic resonance motion was located within a small region on the ventral bank imaging (fMRI) to reveal neural activity uniquely associated of the occipital extent of the superior-temporal sulcus (STS). with perception of biological motion. We isolated brain areas This region is located lateral and anterior to human MT/MST, activated during the viewing of point-light figures, then and anterior to KO. Among our observers, we localized this compared those areas to regions known to be involved in region more frequently in the right hemisphere than in the coherent-motion perception and kinetic-boundary perception. left. This was true regardless of whether the point-light figures Coherent motion activated a region matching previous reports were presented in the right or left hemifield. A small region of human MT/MST complex located on the temporo-parieto- in the medial cerebellum was also active when observers occipital junction. Kinetic boundaries activated a region viewed biological-motion sequences. Consistent with earlier posterior and adjacent to human MT previously identified as neuroimaging and single-unit studies, this pattern of results the kinetic-occipital (KO) region or the lateral-occipital (LO) points to the existence of neural mechanisms specialized complex. The pattern of activation during viewing of biological for analysis of the kinematics defining biological motion. & INTRODUCTION Of course, motion information is important for more We live in a dynamic visual world: Objects often move than just identification of biological motion. Humans are about in our environment and we, as observers, are adept at detecting weak coherent motion amidst a often moving as we look at those events. One of the background of incoherent motion (Raymond, 1994; most biologically salient events is, in fact, human van de Grind, Koenderink, van Doorn, Milders, & Voer- movement. People are remarkably adept at recognizing man, 1993; Williams & Sekuler, 1984; but see Barlow & the actions performed by others, even when the Tripathy, 1997), are very accurate at judging the direc- kinematic patterns of their movements are portrayed tion in which objects are moving (Gros, Blake, & Hiris, by nothing more than a handful of light points 1998; De Bruyn & Orban, 1988; Pasternak & Merigan, attached to the head and major joints of the body 1984; Ball & Sekuler, 1982), and, under optimal condi- (Johannson, 1973). Individual, static frames look like tions, are keenly sensitive to slight differences in the meaningless clusters of dots, but when the frames are speed at which objects are moving (Chen, Bedell, & animated, one immediately perceives a biological or- Frishman, 1998; McKee & Welch, 1989; De Bruyn & ganism engaged in a readily identifiable activity. With Orban, 1988; Orban, De Wolf, & Maes, 1984). In addi- no more than 12 point-lights portraying biological tion, motion provides a potent source of information for motion, people can reliably discriminate male from specifying the 3-D shapes of objects (Tittle & Perotti, female actors, friends from strangers (Cutting & Ko- 1998; Sperling, Landy, Dosher, & Perkins, 1989; Lappin zlowski, 1977), and even subtle differences in complex & Fuqua, 1983; Rogers & Graham, 1979; Wallach & activities such as serving a tennis ball (Pollick, Fido- O’Connell, 1953). piastis, & Braden, 1999). Perception of biological mo- Reflecting the importance of motion information in tion is also robust to variations in the number of dots perception, the primate visual system devotes consider- used and variations in exposure duration (Neri, Mor- able neural machinery to the analysis of motion informa- rone, & Burr, 1998). Despite its robustness, perception tion. From single-cell studies, it is known that neurons of biological motion is severely degraded simply by specialized for registration of motion are found in multi- inverting the animation sequences, thereby portraying ple visual areas of the dorsal pathways projecting from the activity upside-down (Ahlstrom, Blake, & Ahlstrom, area V1, through V5/MT and to higher visual areas within 1997; Sumi, 1984). the parietal lobe (Andersen, 1997; Newsome, 1997). In © 2000 Massachusetts Institute of Technology Journal of Cognitive Neuroscience 12:5, pp. 711–720 recent years, important advances have been made in tion of biological motion, optic flow, and coherent relating activity of neurons in some of these areas (most motion. notably, MT and MST) to visual registration of coherent- While these three studies agree that biological-motion translating motion (e.g., Shadlen, Britten, Newsome, & perception may be linked to activity within a region near Movshon, 1996), 3-D surface shape from motion parallax the posterior STS, it is questionable from those results (e.g., Bradley, Chang, & Andersen, 1998; Xiao, Marcar, whether this is the only neural area uniquely involved in Raiguel, & Orban, 1997), and complex-optical flow asso- seeing biological motion. In addition, it remained to be ciated with looming and rotation (Lagae, Maes, Xiao, learned where this putative area exists relative to areas Raiguel, & Orban, 1994; Duffy & Wurtz, 1991; Saito et al., activated by other, non-biological animation sequences. 1986). However, neurons in these visual areas do not The present experiments sought to address these re- seem particularly well-suited for the extraction of the maining issues using fMRI. Specifically, we compared complex kinematics characterizing biological motion. visual areas activated by biological motion to brain loci Are there other, more specialized areas where this responsive to coherent motion, a stimulus known to processing might be accomplished? Single-unit record- activate the MT/MST complex (Cheng, Fujita, Kanno, ings in the superior-temporal-polysensory area (STPa) of Miura, & Tanaka, 1995; Tootell et al., 1995; Dupont, macaque monkeys have identified cells that respond Orban, De Bruyn, Verbruggen, & Mortelsman, 1994; selectively to biological motions, including some por- Watson et al., 1993) and to motion-defined shapes, a trayed by point-light animations (Oram & Perrett, 1994). class of stimuli that activates the kinetic-occipital (KO) The STPa, incidentally, receives input from both the region (Van Oostende, Sunaert, Van Hecke, Marchal, & dorsal and ventral pathways, which is befitting the multi- Orban, 1997; Orban et al., 1995) or area LOC/LOP as it is dimensional nature of biological creatures engaged in called by some (Tootell, Hadjikhani, Mendola, Marrett, & diverse activities. Dale, 1998). The inclusion of motion-defined shapes in Is there evidence for the existence of neural me- our study seemed particularly relevant given that chanisms in the human brain specialized for the biological motion has been characterized as a mo- analysis of biological motion? Several lines of evidence tion-contour-discrimination problem (Grossberg & point to an affirmative answer. Schenk and Zihl Rudd, 1989). (1997a, 1997b) have described two patients with normal sensitivity to weak, coherent motion but with impaired ability to detect biological-motion figures RESULTS portrayed against a background of static noise. On the other hand, the well-known patient, L.M., suffers Group Results impaired-speed-discrimination and elevated-coher- Averaging statistical maps for all seven observers in ence-detection thresholds, yet she experiences no stereotaxic space revealed stimulus-specific clusters of difficulty recognizing and describing human activities activation. Here, we consider each of these clusters in portrayed by point-light displays (Vaina, Lemay, Bi- turn. enfang, Choi, & Nakayama, 1990). These dissociated The coherent-motion condition (coherent motion deficits imply that the neural mechanisms involved in minus static dots) revealed bilateral clusters of intense detection of simple-translational motion differ anato- activity within a sulcus between the lateral-occipital mically from those underlying perception of biologi- (LO) sulcus and the inferior-occipital sulcus, at the cal motion. juncture of the temporal, parietal, and occipital lobes Several recent neuroimaging studies also point to (see Figure 1B). Previous reports define this location the existence of a region in the human brain uniquely as the MT/MST complex (Sunaert, Van Hecke, Marchal, activated by biological motion (perhaps the analog to & Orban, 1999; Tootell et al., 1995; Dupont et al., the macaque STPa). Using functional magnetic reso- 1994; Watson et al., 1993). While activation was re- nance imaging (fMRI), Puce, Allison, Bentin, Gore, and liably bilateral, the precise location and extent differed McCarthy (1998) isolated a region in the right-poster- slightly between the two hemispheres, with the right- ior-superior-temporal sulcus (STS) that was activated hemisphere MT+ appearing slightly dorsal and medial when people viewed animation sequences showing relative to the left hemisphere. Others have reported eye and mouth movements. Using positron emission more variability in the left-hemisphere MT+ across tomography (PET), Bonda, Petrides, Ostry, and Evans observers (Howard et al., 1996), and our results also (1996) found that this same general region was acti- show this tendency. vated when observers viewed point-light figures. Also The kinetic-boundary minus coherent-motion condi- using fMRI, Howard et